This is a post write up on passing the Cisco R&S CCIE lab exam, this was my second lab attempt, 
almost 9 months between first and second attempts.

My post written exam study materials were:

-Software-
- Doyles - Routing TCP/IP volumes 1 and 2 - Must Read!
- Odom - Cisco QoS - Must Read!
- Inside Cisco IOS Software Architecture
- Parkhurst - Cisco BGP-4 command & configuration handbook
- Parkhurst - Cisco OSPF command & configuration handbook
- Desmeules - Implementing Cisco IPv6 Networks (IPV6) - Must Read!
- Kennedy Clark - Cisco LAN Switching - Must Read!
- www.google.com - is your study partner.
- http://www.cisco.com/univercd - all revelant product documentation.
- www.netmasterclass.com Doit v3 workbook. - These guys rock.
- www.internetworkexpert.com mocklab workbook - emulates real backbone connections in each lab. 

-Hardware-
4x3640-128/32, 2x2621-64/32, 1x3620-64/16, 1x3660-128/64, 4x2500-16/16, 2x7206-256/82, 2x3550-64/16
no ISDN no ATM (1 7206 was frame switch, 1 2511 terminal server)
I tried to get close to the NMC and internetwork export pod configurations.  
(you do not need to upgrade your 2620/1's to 32MB flash for the NMC workbook,  
only the internetworkexpert workbook required this.)

The things I think that were the most helpful in passing the ccie lab were:

1. I think this was the most important. Be completely relaxed durring the
lab exam. You shouldn't be stressed out, that is very counter-productive.
I found doing some checkit labs really helped with this as actually
working within a timed limit teaches you how to manage time. I would
deffently recommend doing atleast one or two checkit labs a few weeks
before your lab date. When you get to the real lab. Take your time to
review all the tasks before even touching the keyboard. Take some notes if
you think one task will affect another task. This part if probably bad
advice, but if you feel comfortable with a certain task and see that other
tasks later in the book add to its configuration (say some kind of
authentication) configure everything for those tasks at once, 
but keep track of what you do. This should save you time barring any typos.

This also means getting a very good nights rest. I myself am a bit of a
insomniac so getting up at 5am was very diffcult. I ended up staying up for
about 28 hours before hand so I was very tired by the time I got to the
hotel and was able to go to sleep around 9pm. The next day for the exam, I
walked in and already knew the process so I was able to keep the stress
down. Also for your home study - setup your terminal someplace that is
free from distractions. I ended up converting my kitchen table into my
study area, I live in a small apartment and the other areas had too many
distractions. I also tried to duplicate the testing environment as close
as possible - for example it is widely known that RTP uses SecureCRT as a
terminal emulator, So I moved from using aterm or putty to SecureCRT
deffently helped me get used to the different copy and paste methods.

2. Part of being relaxed is being comfortable with the material but not
overdoing studying. The first attempt I took, I brought study materials
with me on the airplane and was studying the night before the exam. This
time I stopped all studying 2 days before the exam. Powered down the lab
gear, put away the books and just tried to relax. The only thing I
brought with me to RTP was a change of clothes and some manga to read on
the airplane.

3. Ask the Proctor anything. Anything that is on the exam, and your not
100% sure on what the task is asking - get clarification from the proctor.
Thats what they are there for. I must have asked atleast 10 questions for
clarification. One question I even asked both proctors for different
views. Another I rephrased twice to be absolutely sure. They won't tell
you the answer, but may steer you in the right direction. And only do what
the exam asks for - make no assumptions. I had the most difficulity with
this.

4. Finish Early to check your work. Try to plan on allowing yourself 45-30
minutes for going back and checking your configurations and rebooting the
rack, even if you still have a few 2-3 point questions unfinished. I ended
up finishing up about an hour and half early, and found 2 really stupid
mistakes and typos that I really think would have cost me the exam. Also
running a TCL script was very useful and easy to remember if you use one
for each practice lab. Bascially define a variable, give a dataset to the
variable and then tell it to do something with that set:

foreach ipv4 {
1.1.1.1
..address array..
} {ping $ipv4}

use 'show ip alias' and the terminals 'block copy' function (hold down the
alt-key when highlighting) to get the addresses into notepad it apppears
to copy CR/LFs just fine.

5. Notepad and the windows caculator - Use them! notepad is very useful if
you have alot of duplicate configurations for each device such as
frame-relay class-maps or larger BGP configurations, use notepad as a
place to store and modify this configuration. I've found myself sometimes
doing the confiugration in notepad and then copying to the devices.
The calcuator- If you are converting anything to binary, use the calcuator
convert from different base systems for you. Its too easy to make a
mistake by hand esp when converting many decimal or hex numbers. Just
remember the calculator does not 'pad' to a byte/word. So when copying to
paper read from right to left and pad in the needed MSB's.


6. Enable RIB event debugging. enable 'debug ip routing' - if any routing
information gets entered into the RIB/FIB or if any change occurs this
will let you know whats going on. I found this very helpful during IGP
configuration. Just remember to turn it off if you choose not to reboot
your rack. other useful debugs I've found myself leaving on are 'debug ip
mrouting rpf-events' 'debug ip pim' and 'debug ipv6 routing'. In newer IOS
enabling 'debug ip routing' might cause a router with a default route
learned from  IGP to spew out something about 0.0.0.0/0 every minute, but
this appears to be normal/ok.


7. If you don't do 100% on a practice lab don't let that get you down. The
practice labs are designed to be harder then the actual CCIE lab. It
doesn't appear that they do that just for kicks. (although I've seen a few
I have to question if someone was laughing while writing the task :)

Other small useful things to remember:

for spanning tree remmeber:

DUD
CPI

D=downstream U=upstream (direction C/P/I value affects)
C=cost
P=priority
I=interface #

for ipv6 nat PT in the final configuration you'll see:

ipv4v6  V4 "real address" | V6 translated address
ipv6v4  V6 "real address" | V4 translated address

the real addresses will always be on the left translated on the right.

remember to add a network to any IGP/EGP if it is to announce a summary
route, it needs to be native to the protocol to do so. ie. in EIGRP have a
network statement for a network that is longer or equal to the
summary-address from the interface configuration.

NBMA split-horizon is enabled for all interfaces in EIGRP, auto disabled
on physical interfaces for RIP.

Tc = Bc/CIR

serilization delay = FRAME bits / link bps

BPDU filtering per port is a bad idea. (basically disabled spanning tree)

OSPF cost = reference bandwidth to 8th power / link BW

FRF.12 LFI fragment size = max delay * bandwidth - entered in BYTES

for PPP multilink and PPP authentication, works best if the Virtual
templates contain the authentication configuration.

caculating EIGRP load variance remember a link can only be used if its a
feasible secessor, you can tweak the bandwidth to make this happen.

to calculate EIGRP load varaiance (this is probably a horrible way to do
this):  Sucessor Metric * Configured Variance Value = Target Metric.
Target Metric - (10 to the 7th power / min link BW) = delay/256. set this
value as interface delay subtracting the existing interface delay. This
takes awhile to fully understand, but I would highly recommend spending
the time with this. NMC has a good paper on this also there are some good
GS archive posts on this.

In the lab, the DOCCD search function really isn't good. but If you need
to use it, adding the strings +univercd or +123newft are very useful.

:wq!

-nick #15803